Skip to content
API Reference
Fine Tuning
Jobs

Resume fine-tuning

fine_tuning.jobs.resume(strfine_tuning_job_id) -> FineTuningJob
POST/fine_tuning/jobs/{fine_tuning_job_id}/resume

Resume a fine-tune job.

ParametersExpand Collapse
fine_tuning_job_id: str
ReturnsExpand Collapse
class FineTuningJob: …

The fine_tuning.job object represents a fine-tuning job that has been created through the API.

id: str

The object identifier, which can be referenced in the API endpoints.

created_at: int

The Unix timestamp (in seconds) for when the fine-tuning job was created.

error: Optional[Error]

For fine-tuning jobs that have failed, this will contain more information on the cause of the failure.

code: str

A machine-readable error code.

message: str

A human-readable error message.

param: Optional[str]

The parameter that was invalid, usually training_file or validation_file. This field will be null if the failure was not parameter-specific.

fine_tuned_model: Optional[str]

The name of the fine-tuned model that is being created. The value will be null if the fine-tuning job is still running.

finished_at: Optional[int]

The Unix timestamp (in seconds) for when the fine-tuning job was finished. The value will be null if the fine-tuning job is still running.

hyperparameters: Hyperparameters

The hyperparameters used for the fine-tuning job. This value will only be returned when running supervised jobs.

batch_size: Optional[Union[Literal["auto"], int, null]]

Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.

Accepts one of the following:
Literal["auto"]
int
learning_rate_multiplier: Optional[Union[Literal["auto"], float, null]]

Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.

Accepts one of the following:
Literal["auto"]
float
n_epochs: Optional[Union[Literal["auto"], int, null]]

The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.

Accepts one of the following:
Literal["auto"]
int
model: str

The base model that is being fine-tuned.

object: Literal["fine_tuning.job"]

The object type, which is always "fine_tuning.job".

organization_id: str

The organization that owns the fine-tuning job.

result_files: List[str]

The compiled results file ID(s) for the fine-tuning job. You can retrieve the results with the Files API.

seed: int

The seed used for the fine-tuning job.

status: Literal["validating_files", "queued", "running", 3 more]

The current status of the fine-tuning job, which can be either validating_files, queued, running, succeeded, failed, or cancelled.

Accepts one of the following:
"validating_files"
"queued"
"running"
"succeeded"
"failed"
"cancelled"
trained_tokens: Optional[int]

The total number of billable tokens processed by this fine-tuning job. The value will be null if the fine-tuning job is still running.

training_file: str

The file ID used for training. You can retrieve the training data with the Files API.

validation_file: Optional[str]

The file ID used for validation. You can retrieve the validation results with the Files API.

estimated_finish: Optional[int]

The Unix timestamp (in seconds) for when the fine-tuning job is estimated to finish. The value will be null if the fine-tuning job is not running.

integrations: Optional[List[FineTuningJobWandbIntegrationObject]]

A list of integrations to enable for this fine-tuning job.

type: Literal["wandb"]

The type of the integration being enabled for the fine-tuning job

wandb: FineTuningJobWandbIntegration

The settings for your integration with Weights and Biases. This payload specifies the project that metrics will be sent to. Optionally, you can set an explicit display name for your run, add tags to your run, and set a default entity (team, username, etc) to be associated with your run.

project: str

The name of the project that the new run will be created under.

entity: Optional[str]

The entity to use for the run. This allows you to set the team or username of the WandB user that you would like associated with the run. If not set, the default entity for the registered WandB API key is used.

name: Optional[str]

A display name to set for the run. If not set, we will use the Job ID as the name.

tags: Optional[List[str]]

A list of tags to be attached to the newly created run. These tags are passed through directly to WandB. Some default tags are generated by OpenAI: "openai/finetune", "openai/{base-model}", "openai/{ftjob-abcdef}".

metadata: Optional[Metadata]

Set of 16 key-value pairs that can be attached to an object. This can be useful for storing additional information about the object in a structured format, and querying for objects via API or the dashboard.

Keys are strings with a maximum length of 64 characters. Values are strings with a maximum length of 512 characters.

method: Optional[Method]

The method used for fine-tuning.

type: Literal["supervised", "dpo", "reinforcement"]

The type of method. Is either supervised, dpo, or reinforcement.

Accepts one of the following:
"supervised"
"dpo"
"reinforcement"
dpo: Optional[DpoMethod]

Configuration for the DPO fine-tuning method.

hyperparameters: Optional[DpoHyperparameters]

The hyperparameters used for the DPO fine-tuning job.

batch_size: Optional[Union[Literal["auto"], int, null]]

Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.

Accepts one of the following:
Literal["auto"]
int
beta: Optional[Union[Literal["auto"], float, null]]

The beta value for the DPO method. A higher beta value will increase the weight of the penalty between the policy and reference model.

Accepts one of the following:
Literal["auto"]
float
learning_rate_multiplier: Optional[Union[Literal["auto"], float, null]]

Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.

Accepts one of the following:
Literal["auto"]
float
n_epochs: Optional[Union[Literal["auto"], int, null]]

The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.

Accepts one of the following:
Literal["auto"]
int
reinforcement: Optional[ReinforcementMethod]

Configuration for the reinforcement fine-tuning method.

grader: Grader

The grader used for the fine-tuning job.

Accepts one of the following:
class StringCheckGrader: …

A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.

input: str

The input text. This may include template strings.

name: str

The name of the grader.

operation: Literal["eq", "ne", "like", "ilike"]

The string check operation to perform. One of eq, ne, like, or ilike.

Accepts one of the following:
"eq"
"ne"
"like"
"ilike"
reference: str

The reference text. This may include template strings.

type: Literal["string_check"]

The object type, which is always string_check.

class TextSimilarityGrader: …

A TextSimilarityGrader object which grades text based on similarity metrics.

evaluation_metric: Literal["cosine", "fuzzy_match", "bleu", 8 more]

The evaluation metric to use. One of cosine, fuzzy_match, bleu, gleu, meteor, rouge_1, rouge_2, rouge_3, rouge_4, rouge_5, or rouge_l.

Accepts one of the following:
"cosine"
"fuzzy_match"
"bleu"
"gleu"
"meteor"
"rouge_1"
"rouge_2"
"rouge_3"
"rouge_4"
"rouge_5"
"rouge_l"
input: str

The text being graded.

name: str

The name of the grader.

reference: str

The text being graded against.

type: Literal["text_similarity"]

The type of grader.

class PythonGrader: …

A PythonGrader object that runs a python script on the input.

name: str

The name of the grader.

source: str

The source code of the python script.

type: Literal["python"]

The object type, which is always python.

image_tag: Optional[str]

The image tag to use for the python script.

class ScoreModelGrader: …

A ScoreModelGrader object that uses a model to assign a score to the input.

input: List[Input]

The input messages evaluated by the grader. Supports text, output text, input image, and input audio content blocks, and may include template strings.

content: InputContent

Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.

Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class InputContentOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class InputContentInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

List[GraderInputItem]
Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class GraderInputItemOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class GraderInputItemInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

role: Literal["user", "assistant", "system", "developer"]

The role of the message input. One of user, assistant, system, or developer.

Accepts one of the following:
"user"
"assistant"
"system"
"developer"
type: Optional[Literal["message"]]

The type of the message input. Always message.

model: str

The model to use for the evaluation.

name: str

The name of the grader.

type: Literal["score_model"]

The object type, which is always score_model.

range: Optional[List[float]]

The range of the score. Defaults to [0, 1].

sampling_params: Optional[SamplingParams]

The sampling parameters for the model.

max_completions_tokens: Optional[int]

The maximum number of tokens the grader model may generate in its response.

minimum1
reasoning_effort: Optional[ReasoningEffort]

Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response.

  • gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1.
  • All models before gpt-5.1 default to medium reasoning effort, and do not support none.
  • The gpt-5-pro model defaults to (and only supports) high reasoning effort.
  • xhigh is supported for all models after gpt-5.1-codex-max.
Accepts one of the following:
"none"
"minimal"
"low"
"medium"
"high"
"xhigh"
seed: Optional[int]

A seed value to initialize the randomness, during sampling.

temperature: Optional[float]

A higher temperature increases randomness in the outputs.

top_p: Optional[float]

An alternative to temperature for nucleus sampling; 1.0 includes all tokens.

class MultiGrader: …

A MultiGrader object combines the output of multiple graders to produce a single score.

calculate_output: str

A formula to calculate the output based on grader results.

graders: Graders

A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.

Accepts one of the following:
class StringCheckGrader: …

A StringCheckGrader object that performs a string comparison between input and reference using a specified operation.

input: str

The input text. This may include template strings.

name: str

The name of the grader.

operation: Literal["eq", "ne", "like", "ilike"]

The string check operation to perform. One of eq, ne, like, or ilike.

Accepts one of the following:
"eq"
"ne"
"like"
"ilike"
reference: str

The reference text. This may include template strings.

type: Literal["string_check"]

The object type, which is always string_check.

class TextSimilarityGrader: …

A TextSimilarityGrader object which grades text based on similarity metrics.

evaluation_metric: Literal["cosine", "fuzzy_match", "bleu", 8 more]

The evaluation metric to use. One of cosine, fuzzy_match, bleu, gleu, meteor, rouge_1, rouge_2, rouge_3, rouge_4, rouge_5, or rouge_l.

Accepts one of the following:
"cosine"
"fuzzy_match"
"bleu"
"gleu"
"meteor"
"rouge_1"
"rouge_2"
"rouge_3"
"rouge_4"
"rouge_5"
"rouge_l"
input: str

The text being graded.

name: str

The name of the grader.

reference: str

The text being graded against.

type: Literal["text_similarity"]

The type of grader.

class PythonGrader: …

A PythonGrader object that runs a python script on the input.

name: str

The name of the grader.

source: str

The source code of the python script.

type: Literal["python"]

The object type, which is always python.

image_tag: Optional[str]

The image tag to use for the python script.

class ScoreModelGrader: …

A ScoreModelGrader object that uses a model to assign a score to the input.

input: List[Input]

The input messages evaluated by the grader. Supports text, output text, input image, and input audio content blocks, and may include template strings.

content: InputContent

Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.

Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class InputContentOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class InputContentInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

List[GraderInputItem]
Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class GraderInputItemOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class GraderInputItemInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

role: Literal["user", "assistant", "system", "developer"]

The role of the message input. One of user, assistant, system, or developer.

Accepts one of the following:
"user"
"assistant"
"system"
"developer"
type: Optional[Literal["message"]]

The type of the message input. Always message.

model: str

The model to use for the evaluation.

name: str

The name of the grader.

type: Literal["score_model"]

The object type, which is always score_model.

range: Optional[List[float]]

The range of the score. Defaults to [0, 1].

sampling_params: Optional[SamplingParams]

The sampling parameters for the model.

max_completions_tokens: Optional[int]

The maximum number of tokens the grader model may generate in its response.

minimum1
reasoning_effort: Optional[ReasoningEffort]

Constrains effort on reasoning for reasoning models. Currently supported values are none, minimal, low, medium, high, and xhigh. Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response.

  • gpt-5.1 defaults to none, which does not perform reasoning. The supported reasoning values for gpt-5.1 are none, low, medium, and high. Tool calls are supported for all reasoning values in gpt-5.1.
  • All models before gpt-5.1 default to medium reasoning effort, and do not support none.
  • The gpt-5-pro model defaults to (and only supports) high reasoning effort.
  • xhigh is supported for all models after gpt-5.1-codex-max.
Accepts one of the following:
"none"
"minimal"
"low"
"medium"
"high"
"xhigh"
seed: Optional[int]

A seed value to initialize the randomness, during sampling.

temperature: Optional[float]

A higher temperature increases randomness in the outputs.

top_p: Optional[float]

An alternative to temperature for nucleus sampling; 1.0 includes all tokens.

class LabelModelGrader: …

A LabelModelGrader object which uses a model to assign labels to each item in the evaluation.

input: List[Input]
content: InputContent

Inputs to the model - can contain template strings. Supports text, output text, input images, and input audio, either as a single item or an array of items.

Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class InputContentOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class InputContentInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

List[GraderInputItem]
Accepts one of the following:
str

A text input to the model.

class ResponseInputText: …

A text input to the model.

text: str

The text input to the model.

type: Literal["input_text"]

The type of the input item. Always input_text.

class GraderInputItemOutputText: …

A text output from the model.

text: str

The text output from the model.

type: Literal["output_text"]

The type of the output text. Always output_text.

class GraderInputItemInputImage: …

An image input block used within EvalItem content arrays.

image_url: str

The URL of the image input.

type: Literal["input_image"]

The type of the image input. Always input_image.

detail: Optional[str]

The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.

class ResponseInputAudio: …

An audio input to the model.

input_audio: InputAudio
data: str

Base64-encoded audio data.

format: Literal["mp3", "wav"]

The format of the audio data. Currently supported formats are mp3 and wav.

Accepts one of the following:
"mp3"
"wav"
type: Literal["input_audio"]

The type of the input item. Always input_audio.

role: Literal["user", "assistant", "system", "developer"]

The role of the message input. One of user, assistant, system, or developer.

Accepts one of the following:
"user"
"assistant"
"system"
"developer"
type: Optional[Literal["message"]]

The type of the message input. Always message.

labels: List[str]

The labels to assign to each item in the evaluation.

model: str

The model to use for the evaluation. Must support structured outputs.

name: str

The name of the grader.

passing_labels: List[str]

The labels that indicate a passing result. Must be a subset of labels.

type: Literal["label_model"]

The object type, which is always label_model.

name: str

The name of the grader.

type: Literal["multi"]

The object type, which is always multi.

hyperparameters: Optional[ReinforcementHyperparameters]

The hyperparameters used for the reinforcement fine-tuning job.

batch_size: Optional[Union[Literal["auto"], int, null]]

Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.

Accepts one of the following:
Literal["auto"]
int
compute_multiplier: Optional[Union[Literal["auto"], float, null]]

Multiplier on amount of compute used for exploring search space during training.

Accepts one of the following:
Literal["auto"]
float
eval_interval: Optional[Union[Literal["auto"], int, null]]

The number of training steps between evaluation runs.

Accepts one of the following:
Literal["auto"]
int
eval_samples: Optional[Union[Literal["auto"], int, null]]

Number of evaluation samples to generate per training step.

Accepts one of the following:
Literal["auto"]
int
learning_rate_multiplier: Optional[Union[Literal["auto"], float, null]]

Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.

Accepts one of the following:
Literal["auto"]
float
n_epochs: Optional[Union[Literal["auto"], int, null]]

The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.

Accepts one of the following:
Literal["auto"]
int
reasoning_effort: Optional[Literal["default", "low", "medium", "high"]]

Level of reasoning effort.

Accepts one of the following:
"default"
"low"
"medium"
"high"
supervised: Optional[SupervisedMethod]

Configuration for the supervised fine-tuning method.

hyperparameters: Optional[SupervisedHyperparameters]

The hyperparameters used for the fine-tuning job.

batch_size: Optional[Union[Literal["auto"], int, null]]

Number of examples in each batch. A larger batch size means that model parameters are updated less frequently, but with lower variance.

Accepts one of the following:
Literal["auto"]
int
learning_rate_multiplier: Optional[Union[Literal["auto"], float, null]]

Scaling factor for the learning rate. A smaller learning rate may be useful to avoid overfitting.

Accepts one of the following:
Literal["auto"]
float
n_epochs: Optional[Union[Literal["auto"], int, null]]

The number of epochs to train the model for. An epoch refers to one full cycle through the training dataset.

Accepts one of the following:
Literal["auto"]
int

Resume fine-tuning

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ.get("OPENAI_API_KEY"),  # This is the default and can be omitted
)
fine_tuning_job = client.fine_tuning.jobs.resume(
    "ft-AF1WoRqd3aJAHsqc9NY7iL8F",
)
print(fine_tuning_job.id)
{
  "id": "id",
  "created_at": 0,
  "error": {
    "code": "code",
    "message": "message",
    "param": "param"
  },
  "fine_tuned_model": "fine_tuned_model",
  "finished_at": 0,
  "hyperparameters": {
    "batch_size": "auto",
    "learning_rate_multiplier": "auto",
    "n_epochs": "auto"
  },
  "model": "model",
  "object": "fine_tuning.job",
  "organization_id": "organization_id",
  "result_files": [
    "file-abc123"
  ],
  "seed": 0,
  "status": "validating_files",
  "trained_tokens": 0,
  "training_file": "training_file",
  "validation_file": "validation_file",
  "estimated_finish": 0,
  "integrations": [
    {
      "type": "wandb",
      "wandb": {
        "project": "my-wandb-project",
        "entity": "entity",
        "name": "name",
        "tags": [
          "custom-tag"
        ]
      }
    }
  ],
  "metadata": {
    "foo": "string"
  },
  "method": {
    "type": "supervised",
    "dpo": {
      "hyperparameters": {
        "batch_size": "auto",
        "beta": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto"
      }
    },
    "reinforcement": {
      "grader": {
        "input": "input",
        "name": "name",
        "operation": "eq",
        "reference": "reference",
        "type": "string_check"
      },
      "hyperparameters": {
        "batch_size": "auto",
        "compute_multiplier": "auto",
        "eval_interval": "auto",
        "eval_samples": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto",
        "reasoning_effort": "default"
      }
    },
    "supervised": {
      "hyperparameters": {
        "batch_size": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto"
      }
    }
  }
}
Returns Examples
{
  "id": "id",
  "created_at": 0,
  "error": {
    "code": "code",
    "message": "message",
    "param": "param"
  },
  "fine_tuned_model": "fine_tuned_model",
  "finished_at": 0,
  "hyperparameters": {
    "batch_size": "auto",
    "learning_rate_multiplier": "auto",
    "n_epochs": "auto"
  },
  "model": "model",
  "object": "fine_tuning.job",
  "organization_id": "organization_id",
  "result_files": [
    "file-abc123"
  ],
  "seed": 0,
  "status": "validating_files",
  "trained_tokens": 0,
  "training_file": "training_file",
  "validation_file": "validation_file",
  "estimated_finish": 0,
  "integrations": [
    {
      "type": "wandb",
      "wandb": {
        "project": "my-wandb-project",
        "entity": "entity",
        "name": "name",
        "tags": [
          "custom-tag"
        ]
      }
    }
  ],
  "metadata": {
    "foo": "string"
  },
  "method": {
    "type": "supervised",
    "dpo": {
      "hyperparameters": {
        "batch_size": "auto",
        "beta": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto"
      }
    },
    "reinforcement": {
      "grader": {
        "input": "input",
        "name": "name",
        "operation": "eq",
        "reference": "reference",
        "type": "string_check"
      },
      "hyperparameters": {
        "batch_size": "auto",
        "compute_multiplier": "auto",
        "eval_interval": "auto",
        "eval_samples": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto",
        "reasoning_effort": "default"
      }
    },
    "supervised": {
      "hyperparameters": {
        "batch_size": "auto",
        "learning_rate_multiplier": "auto",
        "n_epochs": "auto"
      }
    }
  }
}